Download Traumatic Pneumothorax Detection with Thoracic US

Radiology
Kevin R. Rowan, MD
Andrew W. Kirkpatrick,
MD, FRCSC
David Liu, MD
Kevin E. Forkheim, MD
John R. Mayo, MD
Savvas Nicolaou, MD,
FRCPC
Index terms:
Pneumothorax, 60.732
Thorax, CT, 60.12111, 60.12112,
60.12115
Thorax, radiography, 60.11
Thorax, US, 60.12981, 60.12989
Published online before print
10.1148/radiol.2251011102
Radiology 2002; 225:210 –214
Abbreviation:
FAST ⫽ focused assessment with
sonography in trauma
1
From the Department of Radiology
(K.R.R., D.L., K.E.F., J.R.M., S.N.) and
Section of Trauma Services (A.W.K.),
Vancouver Hospital and Health Sciences Centre, 899 W 12th Ave, Vancouver, British Columbia, Canada V5Z
1M9. From the 2000 RSNA scientific
assembly. Received June 25, 2001; revision requested July 27; final revision
received March 20, 2002; accepted
April 11. Address correspondence to
S.N. (e-mail: [email protected]).
Traumatic Pneumothorax
Detection with Thoracic US:
Correlation with Chest
Radiography and CT—Initial
Experience1
PURPOSE: To prospectively compare the accuracy of ultrasonography (US) with
that of supine chest radiography in the detection of traumatic pneumothoraces,
with computed tomography (CT) as the reference standard.
MATERIALS AND METHODS: Thoracic US, supine chest radiography, and CT
were performed to assess for pneumothorax in 27 patients who sustained blunt
thoracic trauma. US and radiographic findings were compared with CT findings, the
reference standard, for pneumothorax detection. For the purpose of this study, the
sonographers were blinded to the radiographic and CT findings.
RESULTS: Eleven of 27 patients had pneumothorax at CT. All 11 of these pneumothoraces were detected at US, and four were seen at supine chest radiography. In
the one false-positive US case, the patient was shown to have substantial bullous
emphysema at CT. Sensitivity and negative predictive value of US were 100% (11 of
11 and 15 of 15 patients, respectively), specificity was 94% (15 of 16 patients), and
positive predictive value was 92% (11 of 12 patients). Chest radiography had 36%
(four of 11 patients) sensitivity, 100% (16 of 16 patients) specificity, a 100% (four
of four patients) positive predictive value, and a 70% (16 of 23 patients) negative
predictive value.
CONCLUSION: In this study, US was more sensitive than supine chest radiography
and as sensitive as CT in the detection of traumatic pneumothoraces.
©
RSNA, 2002
Supplemental material: radiology.rsnajnls.org/cgi/content/full/2251011102/DC1.
Author contributions:
Guarantors of integrity of entire study,
S.N., J.R.M.; study concepts and design, all authors; literature research,
S.N., K.R.R.; clinical studies, K.R.R.,
S.N.; data acquisition and analysis/interpretation, all authors; statistical
analysis, all authors; manuscript preparation, all authors; manuscript definition of intellectual content, J.R.M.,
K.R.R., S.N.; manuscript editing, revision/review, and final version approval, all authors.
©
RSNA, 2002
210
Patients who sustain trauma (hereafter referred to as trauma patients) to the thorax are at
risk of serious morbidity and mortality from several forms of injury. One of the most
common—and most easily treated—injuries is pneumothorax. Pneumothoraces often are
detected by means of a combination of clinical examination and chest radiography.
Although these techniques are reliable for the detection of large pneumothoraces, a subtle
pneumothorax may be difficult to detect in a trauma situation for several reasons. Pneumothorax may not be clinically evident if it does not cause substantial respiratory compromise or if it causes only a subtle decrease in air entry, which may not be detectable at
auscultation.
In addition, the chest radiographs obtained in trauma settings usually are anteroposterior images obtained with the patient in the supine position, and a pneumothorax may
not be evident if it does not produce a deep sulcus sign, sharp delineation of the pericardial
silhouette, or large asymmetric area of hyperlucency in one of the hemithoraces (1,2).
Radiographs obtained with the patient upright are substantially more sensitive for depicting pneumothoraces, but most trauma patients cannot be positioned upright because of
competing concerns, such as cervical spine precautions, hemodynamic instability, immobilization of orthopedic injuries, ongoing resuscitation, and/or decreased level of con-
Radiology
Figure 1. Longitudinal US image of a normal
anterior intercostal space in a 25-year-old man,
visualized by applying a 7-MHz linear probe.
The arrow points to the horizontal hyperechogenic line that represents the pleural–thoracic
wall interface, and the arrowhead points to the
vertical comet-tail artifact. At real-time US imaging, to-and-fro movement, representing
lung sliding, would be seen at this interface.
The lung sliding and comet-tail artifacts are
synchronized with respiratory movement.
sciousness. Computed tomography (CT)
is the reference standard for the detection of pneumothoraces, because it is the
most sensitive and specific modality in
this clinical setting (3,4). Trauma patients undergo thoracic CT for various
indications, including possible aortic,
lung parenchymal, or thoracic spinal injuries. Although CT is the reference standard, it is neither practical nor feasible to
perform this imaging examination in all
trauma patients to rule out pneumothorax.
The focused assessment with sonography in trauma (FAST) examination has
gained acceptance as a rapid and accurate
tool to screen for free intraperitoneal
fluid associated with visceral injury and
has almost completely replaced diagnostic peritoneal lavage as the initial screening examination in most major trauma
centers (5,6). Ultrasonography (US) is
also commonly used to detect pleural or
pericardial fluid in trauma patients. Techniques in which US was used to detect
pneumothoraces have been described
(3,7,8).
US does not enable visualization of the
entire lung because of the high acoustic
impedance of air (9). However, a highfrequency linear US probe applied to an
Volume 225
䡠
Number 1
intercostal space enables visualization of
the echogenic interface between the
chest wall soft tissues and the aerated
lung. To-and-fro movement of the visceral and parietal pleural surfaces at this
interface can be seen at US imaging as the
patient respires, and this motion is
termed lung sliding (10). A second US feature of the pleural interface is comet-tail
artifacts (3,7), which are hyperechoic reverberation artifacts that extend from the
pleural interface to the distal edge of the
US image (Fig 1; Movie 1, radiology.rsnajnls
.org/cgi/content/full/2251011102/DC1).
The presence of lung sliding and comettail artifacts at the pleural interface indicates apposition of the pleural surfaces.
These US features are absent when the
pleural surfaces are separated by air
within the pleural space. These US findings have been shown to be accurate and
sensitive for the diagnosis of non–traumainduced pneumothorax (8,10,11). This
high-frequency-probe US technique was
recently described as a rapid and effective
technique for trauma patients in Detroit,
Mich, although the US findings were not
correlated with CT findings in these reports (12). The purpose of this study was
to prospectively compare the accuracy of
US with that of supine chest radiography
in the detection of traumatic pneumothoraces, with CT as the reference standard.
MATERIALS AND METHODS
We performed a prospective blinded
study in which patients who had sustained blunt thoracic trauma were examined for pneumothorax at thoracic US
and supine chest radiography. Patients
were eligible for inclusion in the study if,
in the opinion of the attending emergency
physician or trauma surgeon (A.W.K.),
chest imaging was warranted. Most of the
study patients met the criteria for trauma
team activation, which at our institution
include a respiratory rate of less than 10
or greater than 29 breaths per minute, a
systolic blood pressure of less than 90
mm Hg, a Glasgow Coma Scale grade of
less than or equal to 13, and anatomic
injuries associated with a substantial
mechanism or a life-threatening condition (13). Patients in respiratory distress
who were clinically suspected of having
pneumothorax were treated for such
with thoracostomy tube placement, in
accordance with current clinical practice
guidelines (14). The patients treated with
tube thoracostomy prior to imaging were
excluded from the study. During 8
months (March to November 2000) of
data collection, 70 patients who presented to the emergency department met
the study entry criteria. Twenty-seven of
these 70 patients also underwent thoracic CT, and they formed our study population.
Thoracic US was performed by either a
staff radiologist (S.N.) or a radiology resident (K.R.R., D.L., K.E.F.) who was
trained in US pneumothorax detection.
All included patients had substantial injuries or a mechanism of injury such that
the referring physician requested a FAST
examination to rule out the presence of
free intraperitoneal fluid. Thoracic US
was performed immediately after abdominal US. The 27 patients (25 male patients, two female patients; mean age, 42
years; age range, 17– 83 years) included in
this study were those who needed to undergo thoracic CT during the course of
the study for standard clinical indications, such as suspicion of blunt thoracic
aortic disruption, detection or clarification of spinal column injury, discordant
findings between chest radiography and
clinical examination, work-up of mediastinal hematoma, and thoracic parenchymal injury. Institutional review board
approval was obtained prior to commencement of the study. Informed consent was
waived as part of the institutional review
board approval, because obtaining informed consent is a standard protocol for
clinically indicated diagnostic procedures
in the trauma setting.
Thoracic US was performed by using a
US imaging unit (model 128XP; Acuson,
Mountain View, Calif) with a 7-MHz linear probe. All patients were in the supine
position during the examination. Bilateral pleural interfaces were examined at
the second to fourth intercostal spaces
anteriorly and at the sixth to eighth
spaces in the midaxillary line. The presence of pneumothorax was determined
by using accepted US criteria—namely,
disappearance of lung sliding and loss of
the comet-tail artifact at the pleural interface. A US-based diagnosis was made
by the sonographer at the time of the
examination, without prior knowledge of
the radiographic findings and before CT
was performed. The US studies obtained
by radiology residents were recorded on
videotape and later reviewed by the staff
radiologist.
In all cases, chest radiographs were obtained with patients in the supine anteroposterior position. Although both upright and lateral decubitus radiographs
are more sensitive for pneumothorax detection than supine radiographs (2), su-
Thoracic US for Detection of Traumatic Pneumothoraces
䡠
211
Radiology
pine radiographs were obtained for various reasons, including decreased level of
consciousness, cervical spine precautions,
orthopedic injuries, and hemodynamic instability. The criteria for radiographic
pneumothorax detection included visualization of the visceral pleura separated
from the chest wall with loss of lung markings laterally, demonstration of a deep
sulcus sign, crisp definition of the hemidiaphragm, and demonstration of a continuous diaphragm sign (2). US examinations generally were performed within
minutes after the chest radiographic acquisitions, with the longest delay being
less than 30 minutes.
CT was performed by using a CT scanning unit (CT HighSpeed Advantage; GE
Medical Systems, Milwaukee, Wis) and
intravenous (ioversol injection 68% [Optiray 320]; Mallinckrodt Canada, PointeClaire, Quebec, Canada) and oral (diatrizoate meglumine 66%, diatrizoate sodium
10% [Gastrografin]; Bracco Diagnostics
Canada, Mississauga, Ontario, Canada)
contrast materials, when indicated, for
evaluation of chest or abdominal trauma.
Helical sections were obtained at a pitch
of 1.5 and reconstructed to 3-mm-thick
sections. CT images were printed on mediastinal and lung windows. The CT criteria for a diagnosis of pneumothorax included any air collections that were
displacing the visceral pleura from the
chest wall (2). Radiographs and CT images were interpreted by staff radiologists
without knowledge of the US findings.
Final dictated reports were reviewed by
the authors (K.R.R., S.N., D.L.) and compared with the radiographs and CT images for verification.
Estimates of sensitivity, specificity, positive predictive value, negative predictive
value, and overall accuracy were calculated
for US versus CT and for radiography versus CT by using CT as the reference standard for pneumothorax detection. Ninetyfive percent CIs based on binomial distribution were calculated for all of the estimates by using a statistical software program (S-Plus 2000; MathSoft, Seattle,
Wash). Logistic regression analysis was used
to evaluate how well the CT results could be
predicted with US and chest radiography.
With use of additional statistical software
(SPSS 10; SPSS, Chicago, Ill), the CT results were treated as a response variable
and US and chest radiography were treated
as two independent predictor variables.
RESULTS
The CT, US, and chest radiographic findings in the 27 patients are compared in
212
䡠
Radiology
䡠
October 2002
TABLE 1
Pneumothorax Depiction at CT, US,
and Chest Radiography
No. of
Patients*
CT
US
Chest
Radiography
15
1
7
4
Absent
Absent
Present
Present
Absent
Present
Present
Present
Absent
Absent
Absent
Present
* Numbers of patients in whom a pneumothorax was present or absent at the given
imaging examination.
Table 1. Eleven of the 27 patients had
pneumothoraces at CT. US demonstrated
the presence of pneumothorax in all 11
of these patients. There were no falsenegative US results. A single false-positive
US case was reported: The absence of
lung sliding in the anterior aspect of one
hemithorax suggested pneumothorax.
The CT scan did not reveal pneumothorax but rather a substantial bullous emphysema with a large anterior bulla in
the area of the US abnormality.
A static (ie, non–real-time) thoracic US
image obtained in a patient who sustained blunt thoracic trauma is shown in
Figure 2. Comet-tail artifact is absent on
this image of the left anterior hemithorax.
Real-time US imaging depicted loss of lung
sliding at this site (Movie 2, radiology.rsnajnls
.org/cgi/content/full/2251011102/DC1). A supine anteroposterior chest radiograph obtained in the same patient is shown in Figure 3. Although there is a suggestion of
increased definition of the left hemidiaphragm on this image, there is no definitive evidence of pneumothorax. The selected thoracic CT image shown in Figure
4, which was obtained in the same patient,
shows a moderate left pneumothorax.
Mild left subcutaneous emphysema, a mild
degree of left lower lobe atelectasis, and a
small left pleural effusion also are present.
Four of the 11 CT-confirmed pneumothoraces were detected at supine radiography. The chest radiographs obtained in
seven of the 11 patients with CT-confirmed pneumothorax showed no definitive evidence of pneumothorax.
Data on the performance of radiography versus CT and on the performance of
US versus CT are summarized in Table 2.
Logistic regression analysis revealed a significant correlation between US and CT
with regard to diagnostic accuracy, even
after chest radiographic findings were
taken into account (logistic regression
P ⬍ .001), but not between radiography
and CT after US findings were taken into
account (logistic regression P ⫽ .35).
Figure 2. Pneumothorax in a 25-year-old
man. Longitudinal US image of the fourth left
anterior intercostal space visualized by applying a 7-MHz linear probe. The arrow points to
a motionless lung–thoracic wall interface that
is devoid of comet-tail artifacts. The absence of
lung sliding is apparent only at real-time US
imaging.
DISCUSSION
Pneumothorax is a serious potential consequence of blunt thoracic trauma (4)
and a potential complication of vascular
access procedures (11). In cases in which
a pneumothorax is clinically evident or is
causing substantial respiratory compromise, treatment is often initiated without
performing imaging (14). In many cases,
however, pneumothoraces may not be
detected at clinical examination. Furthermore, the chest radiographs initially obtained in serious trauma cases are almost
invariably anteroposterior supine images, which can be insensitive for detection of subtle pneumothoraces. Although
a subtle pneumothorax may not be clinically important initially, in certain situations there can be serious implications if
the pneumothorax is not detected. For
example, a pneumothorax can increase
in size if the patient is exposed to decreased atmospheric pressure during air
transport or requires intubation and positive pressure ventilation (15,16).
An accurate, rapid, and practical method
to rule out small pneumothoraces that
could be used repeatedly throughout the
clinical course of a trauma case without
additional radiation exposure would have
important clinical implications for trauma
patient care (17). Several groups of investigators (3,7,8) have found that thoracic US
Rowan et al
Radiology
Figure 3. Supine anteroposterior chest radiograph obtained in the
same patient as in Figure 2 shows left subcutaneous emphysema
(arrowheads). Although no distinct evidence of pneumothorax is
visible, there is a suggestion of sharp definition of the left hemidiaphragm, which could lead to the suspicion of pneumothorax. Further
imaging is required to make a definitive diagnosis.
Figure 4. Transverse 3-mm contrast material– enhanced spiral thoracic CT image obtained in the same patient as in Figures 2 and 3
shows a moderate-sized left pneumothorax (arrowheads). Left subcutaneous emphysema (arrow), a mild degree of left lower lobe atelectasis, and a small left pleural effusion (ⴱ) also are present.
TABLE 2
CIs for Chest Radiography versus CT and for US versus CT for Detection
of Pneumothorax
Chest Radiography
US
Parameter
Estimate
95% CI
Estimate
95% CI
Sensitivity (%)
Specificity (%)
False-positive rate (%)
False-negative rate (%)
Positive predictive value (%)
Negative predictive value (%)
Accuracy (%)
Prevalence (%)
36 (4/11)
100 (16/16)
0 (0/16)
64 (7/11)
100 (4/4)
70 (16/23)
74 (20/27)
41 (11/27)
15, 65
81, 100
0, 19
35, 85
51, 100
49, 84
55, 87
25, 59
100 (11/11)
94 (15/16)
6 (1/16)
0 (0/11)
92 (11/12)
100 (15/15)
96 (26/27)
41 (11/27)
74, 100
72, 99
1, 28
0, 24
65, 99
80, 100
82, 99
25, 59
Note.—Data in parentheses are numbers of patients.
can be used to reliably detect pneumothoraces, which are diagnosed on the basis of
loss of lung sliding and absence of comettail artifacts at the hyperechoic pleural interface.
In this small study of 27 trauma patients, US was more sensitive and accurate than supine chest radiography and
as sensitive as thoracic CT in the detection of pneumothoraces. Comparison of
CIs indicated a significant difference (P ⬍
.05) in sensitivities and false-negative ratios between US and chest radiography.
Although the CIs for negative predictive
value and overall accuracy overlap, the
mean value for these parameters indicates that US had a greater than 20%
superiority. This suggests a clinically relevant difference between US and chest
radiography.
Although both the estimated sensitivity and the estimated negative predictive
Volume 225
䡠
Number 1
value of US were 100% in the study
group, the specificity was 94%. These results were due to one false-positive US
case, in which a patient with bullous emphysema showed absence of lung sliding
in the area of a large anterior bulla. This
finding is one of several possible pitfalls
that sonographers should be aware of
when performing US to diagnose pneumothorax. Another potential pitfall is
the presence of pleural adhesions or any
condition in which the pleural surfaces
do not slide against each other at respiration. Additionally, extensive subcutaneous emphysema can interfere with US
visualization of the pleural surfaces. Several of our study patients had small areas
of subcutaneous emphysema; however,
we did not experience difficulty in avoiding the emphysematous collections and
visualizing the pleural surfaces in these
cases.
The sample size in this study was small,
and our inclusion criteria may have introduced bias to the study in two ways:
First, CT was performed in patients for
various clinical indications, such as abnormal mediastinal contour on chest radiographs that raised suspicion of aortic
injury and assessment of lung parenchymal injury. It could be argued that the
patients who required CT, on average,
had more serious injuries than the patients who did not require CT. The second source of bias was that patients were
excluded if they were treated with chest
tube thoracostomy prior to imaging. This
bias, however, possibly lent further credibility to the technique, because patients
with large, clinically important pneumothoraces were excluded, and, therefore,
the pneumothoraces detected at US in
this study were subtle and clinically “silent” at the time of diagnosis. We did not
attempt to estimate the size of the pneumothoraces ultrasonographically, because
previous research has shown that although US is highly sensitive for pneumothorax detection, it does not enable a
reliable estimation of the volume of a
pneumothorax (9). Our study results
demonstrate that US is highly sensitive
and has a high negative predictive value
for the detection of traumatic pneumothorax and therefore can be an effective
diagnostic tool to definitively exclude
pneumothoraces in trauma patients.
The diagnosis of traumatic pneumothorax, especially life-threatening tension
pneumothorax, is usually made on clinical
grounds. There are occasional cases, how-
Thoracic US for Detection of Traumatic Pneumothoraces
䡠
213
Radiology
ever, in which the diagnosis is delayed or
even missed by well-trained residents,
emergency physicians, or trauma specialists (18). Additionally, in settings with lesstrained personnel, limited equipment, or
logistic limitations, even more pneumothoraces will be missed. Newly developed
US equipment includes high-quality portable units that are easily transported by
hand. The use of these units can enable the
assessment of thoracic trauma in many diverse environments, especially situations
in which CT and chest radiography may
not be readily available. Such environments may include patient transport vehicles, conflict or battle zones, commercial
airlines, ocean-going vessels, or manned
space flights. The international space station has already been equipped with an
advanced modified US machine (HDI
5000; Advanced Technology Laboratory,
Bothell, Wash). One of the many potential
applications of this unit is the diagnosis of
pneumothoraces, because recent research
suggests that the techniques used to detect
pneumothorax can also be applied in microgravity conditions (19).
The results of this study suggest that
thoracic US, when performed by trained
individuals, can enable definitive exclusion of pneumothorax. As a corollary, we
believe that thoracic US should be added
to the currently performed FAST examination in trauma cases and be labeled as
214
䡠
Radiology
䡠
October 2002
the expanded FAST examination. With
this new protocol, US would be used to
exclude both intraperitoneal free fluid
and pneumothorax in one focused rapid
assessment of the trauma patient.
10.
11.
References
1. Chan O, Hiorns M. Chest trauma. Eur J
Radiol 1996; 23:23–34.
2. Tocino IM. Pneumothorax in the supine
patient. RadioGraphics 1985; 5:557–586.
3. Goodman TR, Traill ZC, Phillips AJ,
Berger J, Gleeson FV. Ultrasound detection of pneumothorax. Clin Radiol 1999;
54:736 –739.
4. Wall SD, Federle MP, Jeffrey RB, Brett CM.
CT diagnosis of unsuspected pneumothorax after blunt abdominal trauma. AJR
Am J Roentgenol 1983; 141:919 –921.
5. Boulanger BR, Rozycki GS, Rodriguez A.
Sonographic assessment of traumatic injury: future developments. Surg Clin
North Am 1999; 79:1297–1316.
6. Scalea TM, Rodriguez A, Chiu W, et al.
Focused assessment with sonography for
trauma (FAST): results from an international consensus conference. J Trauma
1999; 46:466 – 472.
7. Lichtenstein D, Meziere G, Biderman P,
Gepner A. The comet-tail artifact: an ultrasound sign ruling out pneumothorax.
Intensive Care Med 1999; 25:383–388.
8. Wernecke K, Galanski M, Peters PE, Hansen J. Pneumothorax: evaluation by ultrasound—preliminary results. J Thorac Imaging 1987; 2:76 –78.
9. Sistrom CL, Reiheld CT, Gay SB, Wallace
KK. Detection and estimation of the volume of pneumothorax using real-time
sonography: efficacy determined by re-
12.
13.
14.
15.
16.
17.
18.
19.
ceiver operating characteristic analysis.
AJR Am J Roentgenol 1996; 166:317–321.
Lichenstein DA, Menu Y. A bedside ultrasound sign ruling out pneumothorax in
the critically ill: lung sliding. Chest 1995;
108:1345–1348.
Targhetta R, Bourgeois JM, Chavagneux
R, Balmes P. Diagnosis of pneumothorax
by ultrasound immediately after ultrasonically guided aspiration biopsy. Chest
1992; 101:855– 856.
Dulchavsky SA, Schwarz KL, Kirkpatrick
AW, et al. Prospective evaluation of thoracic ultrasound in the detection of pneumothorax. J Trauma 2001; 50:201–205.
Simons R. Vancouver General Hospital
trauma program manual. 4th ed. Vancouver, British Columbia, Canada: Vancouver General Hospital, 1999.
Esposito TJ, Jurkovich GJ, Rice CL, Maier
RV, Copass MK, Ashbaugh DG. Reappraisal of emergency room thoracotomy
in a changing environment. J Trauma
1991; 31:881– 887.
Baumann MH, Sahn SA. Tension pneumothorax: diagnostic and therapeutic
pitfalls. Crit Care Med 1993; 21:177–179.
Tocino IM, Westcott JL. Barotrauma. Radiol Clin North Am 1996; 34:59 – 81.
Dulchavsky SA, Hamilton DR, Diebel LN,
Sargsyan AE, Billica RD, Williams DR.
Thoracic ultrasound diagnosis of pneumothorax. J Trauma 1999; 47:970 –971.
Kirkpatric AW, Ng A, Dulchavsky SA, et
al. Sonographic diagnosis of a pneumothorax inapparent on plain radiography:
confirmation by computed tomography.
J Trauma 2001; 50:750 –752.
Sargsyan AE, Hamilton DR, Nicolaou S, et
al. Ultrasound evaluation of the magnitude of pneumothorax: a new concept.
Am Surg 2001; 67:232–236.
Rowan et al